Tanmay Debnath on developing AI tools to understand mammalian regulatory networks

Please tell us about your research project, research interests and how your work involves genomics or makes use of genomic technologies 

In my PhD, I am exploring how deep learning and AI could be used as a tool for understanding the regulatory networks of the cattle genome. So the rationale behind my project is to understand how sequence variation could lead to understanding the difference in the expressions of the regulatory elements. So, if we train a model on a better annotated genome like human and try to understand the biology behind the regulatory regions, then using this knowledge, we can understand the variations in the regulatory networks of the cattle genome. This might help us to annotate the cattle genome better without actually having to do experiments (in the future, of course!). I am using massively parallel reporter assay data for this. Furthermore, this tool could also be used as a query machine to understand how evolution might have happened between the respective species.

Please summarise your previous work and career so far.

I started my career as a Mechanical Engineer. My undergraduate study was in that. But during my final year research project, I ventured into AI and developed a software that could detect polyps from colonoscopy videos. Later, I received the prestigious MEXT scholarship to pursue my MS in Japan. I worked with AI and automation at the Toyohashi University of Technology. While in Japan, I developed a platform that could automate the parameters of the photocuring experiment (a system that Prof. Nagai was using to capture single cells) and subsequently perform the experiment without any human intervention, all by itself. If you hear the new buzzword “reinforcement learning” after DeepSeek success, then yes, my method was based on the same principle. Throughout my career, I guess, I have always wanted to extend the boundaries of AI in health and automation.

What inspired you to follow your career path?

Since my childhood, I had an interest in technology and whenever I could find something to tinker with, I got my hands dirty. I remember I completely disassembled my grandma’s radio because I wanted to know how a box can make sounds from thin air. This has helped me throughout my journey so far. I think during Covid times, I pivoted to computational stuff because I realised that the scale at which data is generated by us, is unmatched and there has to be something that can interpret the data and make sense of it. My sensei helped a lot and inspired me to dig deeper into this and explore my computational understanding into real-world projects, more specifically related to health. That really built my confidence and shifted my career in this trajectory.

And the greatest of all, my mom! She taught me that no matter what challenges come, in research or life, if you are adamant enough that you can do it, you can do it! This is why I was able to have an unconventional way of progressing throughout my research career. 

If you could have tea with anyone (alive or dead), who would it be? What would you talk about?

There are so many people I’d like to meet. But I would say, I’d love to have tea with Alan Turing. I’d like to explore his thoughts on breaking Enigma, his views on computers, and how AI (“computers that have brains”) can be explored in almost every domain of human existence. I am very sure we would have such a fun time!

There’s another researcher that I would love to meet. Prof. Deblina Sarkar, from MIT. I think she’s an absolute genius. Her ideas on creating nanodevices that can cross the blood-brain barrier absolutely mind-blowing. Did you know that she designed the first atomically-thin tunnel transistor during her PhD? 

What is the most unusual thing you have done during your career?

The most unusual thing I ever did in my career would be my PhD itself! As I mentioned previously, I am a Mechanical Engineer. And as it could be guessed, nowhere does my previous education lie within the spectrum of biology. From designing gears and products to understanding how transcription factors work, this has been a long journey and more to go. 

Do you have any advice for people wanting to pursue a research career? What do you like best about your job? What do you like the least?

I’d say never stop believing in yourself. Sadly, impostor syndrome is one of the highest in academia. This is the time when one should listen to themselves more. I’d say people possess untapped potential. If they have something of interest and are dedicatedly passionate about it, nothing can stop them – they can take up things they never thought they could. 

The best part about my job will be the opportunity itself. I get to work on something important and the fact that my work has some impact on the real world drives me and motivates me every day.

As much as I love to work on my project, computer engineers don’t necessarily like to get red-flagged errors all around their codes. Every time I try something, my codes just scream errors. This is when I want a time machine so much so that I can just fast-forward time to a point where I solved the error. 

How do you spend your time outside of research? Is there anything else you would like to tell us about? 

I spend most of my time with my friends. We explore new places together and find new ways to annoy each other. Recently, I have started to devote myself to more personal projects. So, working on them would be a priority, hopefully! And, I am a musician myself so I sometimes play with my music production skills and try to make something – cinematically orchestrated, I like to call them!

Links

The Roslin Institute